Skip to main content

Table 2 Performance of SNV and Gene-based approaches using five-fold cross validation

From: Machine learning analysis of exome trios to contrast the genomic architecture of autism and schizophrenia

Method

Accuracy

Precision

Recall

NIR

P-value (Acc > NIR)

95% CI

SNV-based

0.88

0.78

0.97

0.59

< 2.2e-16

(0.86,0.90)

Gene-based

0.88

0.81

0.95

0.57

< 2.2e-16

(0.86,0.90)

  1. The performance between the two algorithms trained to distinguish ASD cases from SCZ cases is measured using five-fold cross validation. All performance metrics are the average of the five cross validation folds
  2. Acc Accuracy, NIR No information rate, CI Confidence interval